This project will examine the effects of a parental marital disruption on the well-being of children. The central question to be addressed is the following: Why does divorce affect some children more than others? About one million American children per year experience a parental divorce, so the proposed topic is very relevant to the health and welfare of children. In this project, divorce is conceptualized as a process that may begin long before the physical separation of the parents and that may take diverse forms after the separation. Consequently, the effects of the divorce process on children's well-being will be examined in three large-scale, longitudinal studies, each of which includes detailed assessments of the well-being of children at two or more points in time, as well as information from the children's parents and other sources. In each study, the subjects to be analyzed will be those children who were in nondisrupted families at the time of the first child assessment. The well-being at later assessments of children whose families subsequently disrupted will then be compared to the well-being of children whose families remained intact. Within the disrupted group, variations in the post-disruption process (such as contact between the noncustodial parent and the child, or the remarriage of the custodial parent) will be examined. Of special note is that all analyses will include indicators that control for observed differences in well-being that already existed at the first (predisruption) assessment. Our statistical analyses will begin with simple regression models and hazard models. Then models that take into account self-selection into the disrupted and nondisrupted groups and sample selection bias will be estimated and evaluated. The three data sets are the National Child Development Study, 1958 Cohort, the Children's Assessment of the National Longitudinal Survey of Youth, and the National Survey of Children.